TECHNICAL NOTE CRIMINALISTICS

Fire investigation is a challenging area for the forensic investigator. The aim of this work was to use spectral changes to paint samples to estimate the temperatures to which a paint has been heated. Five paint samples (one clay paint, two car paints, one metallic paint, and one matt emulsion) have been fully characterized by a combination of attenuated total reflectance Fourier transform infrared (ATR-IR), Raman, X-ray fluorescence spectroscopy and powder X-ray diffraction. The thermal decomposition of these paints has been investigated by means of ATR-IR and thermal gravimetric analysis. Clear temperature markers are observed in the ATR-IR spectra namely: loss of m(C = O) band, >300°C; appearance of water bands on cooling, >500°C; alterations to m(Si–O) bands due to dehydration of silicate clays, >700°C; diminution of m(CO3) and d(CO3) modes of CaCO3, >950°C. We suggest the possible use of portable ATR-IR for nondestructive, in situ analysis of paints.

KEYWORDS: forensic science, paint analysis, infrared spectroscopy, thermogravimetric analysis, fire scene investigation, X-ray diffraction Fire investigation is one of the most challenging studies undertaken by the forensic scientist. The purpose of fire investigation is to determine deliberate ignition, establish liability, and to identify dangerous materials or substances that may have started or contributed to the fire (1). It is an important area-in 2007 alone there were 53,000 dwelling fires and 50,080 road vehicle fires reported within the United Kingdom (1). Paint is an attractive material to study in fire investigation because it is present at most locations where a fire could occur. A major part of fire investigation is to determine the seat of the fire with as much precision as possible and as many as 40 different techniques have been employed to do this (1). An important approach is the "time temperature dependent" technique which assumes that the seat of the fire will have burned for longer and will have reached higher temperatures. Although this assumption may not always be valid, it does, at least, provide a good starting point.
Portable spectroscopic techniques are now becoming much more widely available allowing their use in Scene of Crime investigation for nondestructive in situ analysis. Three such techniques are X-ray fluorescence (XRF), Raman, and attenuated total reflectance Fourier transform infrared (ATR-IR). These techniques allow thorough analysis to be made of paint samples. Crucially both Raman and ATR-IR allow chemical changes to be monitored as a function of temperature. This may allow an estimate of the temperature to which a paint sample has been heated. In this case, the use of these methods may be helpful to the fire scene investigator in identifying the seat of a fire.
Paints are ideal samples to study by ATR-IR. They are typically composed of pigments (2,3) (these may be either inorganic or organic and add color to the paint), resins (to bind the paint together) (4), extenders, and additives (to add bulk to the paint and sometimes to alter its properties) (5,6). All of these components give identifiable IR absorption bands which should change with temperature as the various components decompose. In "wet" paints solvents-either volatile organic compounds (VOCs) (7)-or increasingly water (8) are present but the solvents used will evaporate on heating to leave behind the dried paint. In this work dried paint samples were studied throughout.
The aim of the work reported in this article was to study the changes occurring to a number of paint samples upon heating in the temperature range from room temperature to 1000°C using ATR-IR backed up by thermal gravimetric analysis (TGA). In this way, it was hoped to assess the usefulness of this method for in situ crime scene investigation and to compare the relative merits of ATR-IR, Raman, and XRF spectroscopy in this context.

Experimental Section
All experiments were carried out at The University of Reading. Paint samples were obtained from the suppliers noted in Table 1. For thermal decomposition studies, thin layers of these were painted onto small (25 mm 9 25 mm) ceramic tiles. These were heated in a furnace fitted with a thermostat. Initial characterization of the paint samples was carried out by the following methods. For ATR-IR spectroscopy samples on ceramic tiles were placed in contact with the ATR attachment on a Perkin-Elmer 100 Fourier transform infrared (FTIR) spectrometer 1 (Waltham, MA) and the spectrum was recorded. No further sample preparation was required. Typically, 16 scans were recorded at 2 cm-1 resolution. Raman spectra were similarly recorded from samples on ceramic tiles using a Thermonicolet NXR 9650 FT-Raman spectrometer (Thermo-Scientific, Tewksbury, MA) with a laser wavelength of 1064 nm. Typically, 200 scans were recorded. As such, ATR-IR spectroscopy was preferred over Raman spectroscopy for studying heated samples. Powder X-ray diffraction (XRD) patterns were obtained from dried paint which was scraped from a glass microscope slide and ground to a fine powder. Analysis was carried out using a Bruker D8 Powder diffractometer (Billerica, MA) using monochromatic Cu Ka radiation. XRF was carried out on dried paint on ceramic tiles using a Thermo-Scientific NITON XLt Portable XRF spectrometer. Scanning electron microscopy-energy dispersive X-ray analysis (SEM-EDX) was carried out using an Environmental SEM (FEI, Hillsboro, OR); paint samples were scraped from glass slides and transferred to adhesive metal stubs. It would, however, have been possible-although more time consuming-to study samples directly on ceramic tiles. Analysis was conducted using an Oxford Instruments FEI Quanta FEG 600 Environmental SEM equipped with energy and wavelength dispersive elemental analysis. TGA was carried out on small samples scraped from a glass slide and transferred to a platinum pan of a TA Instruments TGA Q50 instrument (New Castle, DE) operating with a heating range of 10°C/ min over a range from room (ambient) temperature to 1000°C.

Results and Discussion
We chose five paint samples for study. These were selected because they were paints designed for different uses and which are likely to be found at a crime scene and which have different properties: Sample A was a clay-based architectural paint, samples B and C were car paints, B being an automotive undercoat and C an automotive topcoat. D was a metallic paint used for painting metal surfaces with in domestic or industrial sites and E a matt emulsion architectural paint. Table 1 lists the paints studied, their supplier, and the stated and observed colors of the paints. In the first stage of the study, these paints were analyzed by a combination of ATR-IR spectroscopy, Raman spectroscopy, powder XRD, XRF and SEM-EDX. This combination of tech-niques was designed to provide information on elemental composition of inorganic components (XRF and SEM-EDX), mineral content of inorganic components (XRD), and organic and inorganic components (ATR-IR and Raman). Table 2 shows the composition of these paint samples as determined by this combination of techniques. It may be seen that CaCO 3 is a ubiquitous component as an extender. The car paints B and C contain polymethacrylate resin, the others polyvinyl acetate resin. TiO 2 is not present in the car or metallic paints. All samples contain an unidentified organic component which probably result from unidentified organic pigments or (except in the case of the clay paint A) from residual VOC content. The matt emulsion paint E clearly contains the inorganic pigment "Sienna" (Fe 2 O 3 and MgO) whereas in the other paints the color must arise from unidentified organic pigments. Table 3 summarizes the observed pigment, binder, and extender for each sample studied.
All five samples were then heated under controlled conditions from 100°C to 1000°C. Samples of each paint were painted onto small (25 mm 9 25 mm) ceramic tiles and placed in a furnace. The furnace was heated to 100°C for 1 h and one tile was removed for analysis. The furnace was then heated to 150°C for 1 h and the process repeated at 50°C intervals up to a temperature of 1000°C.
Visual inspection of the samples shows two clear color changes. First, between 250°C and 300°C each sample darkens and takes on a brownish color. Then, between 350°C and 400°C the brownish color is lost and the samples become a pale off-white color. Finally, above about 600°C the sample darkens again and the surface structure begins to deteriorate, with the paint sample flaking off. It is of interest that all of these paint samples behaved in a similar way with similar color changes in the same temperature ranges even though they are quite different types of sample.
The results of TGA back up these visual observations. A typical TGA plot is shown for sample C in Fig. 1. Three distinct mass losses are seen: 5.33% between 0°C and 200°C, 34.4% between 250°C and 350°C, and 37.9% between 350°C and 450°C. Table 4 lists the mass losses seen for each of the four samples studied by TGA.
The first mass loss between 200°C and 300°C probably results from decomposition of organic pigments in the paint. These findings would tally with visual inspection of the paint samples where the colors are quite stable up to about 250°C after which  (9)(10)(11). The first leads to loss of acetic acid and the second results from decomposition of the resulting carbon chain. The exact temperatures at which these processes occur are reported to be affected by other components especially metal oxides (12). Polymethacrylates are also reported to decompose in two steps although this is influenced by the exact nature of the polymethacrylate (13). Above about 600°C the darkening of the sample and the degradation of the surface structure corresponds to the final breakdown of the binder. The metallic paint D shows a 5.17% mass loss at low temperatures below 200°C; this probably results from the presence of residual VOCs as the VOC content of this paint was particularly high. The 4.16% mass loss of car paint B above 950°C sample probably results from decomposition of CaCO 3 to CaO and CO 2 -a process which is known to begin around this temperature (14). It may be noted that sample B shows four mass losses whereas samples A, C, and D show only three mass losses. Ideally, and to unambiguously identify the chemical processes giving rise to these mass losses the use of gas chromatography and FTIR could be employed to characterize gases given off upon heating each sample to different temperatures. Such studies were beyond the scope of this work. However, the findings from TGA are backed up by the observed ATR-IR spectra. In Fig. 2 are shown ATR-IR spectra from car paint sample B at room temperature and after heating to 300°C, 500°C, and 700°C. At 300°C the carbonyl group is lost-as a result of loss of acetic acid from the polyvinyl acetate polymer binder. That the organic components are lost in at least two steps is illustrated by the spectra of metallic paint D in Fig. 3 where at 250°C the C-H stretching bands have decayed but the carbonyl absorption at 1731 cm-1 remains even though it has significantly diminished. A very interesting observation from all IR spectra was that upon heating to temperatures above 500°C for 1 h and subsequent cooling, bands of water arising from O-H stretching and bending modes at 3365 cm-1 and 1637 cm-1 appear. After heating to 700°C, these bands become even more prominent (see Fig. 1). The most likely explanation for this is that CaCO 3 is beginning to decompose to form hygroscopic CaO. We note two further observations from the IR spectra. The clay-based paint A shows a prominent υ(Si-O) band at 1005 cm-1. Upon heating to 700°C, this band changes in appearance and the center of the band shifts to 979 cm-1 (see Fig. 4). This results from dehydration of the clay minerals (15). Above 950°C the distinct stretching and bending modes of carbonate at 1415 cm-1 and 874 cm-1 (16) will begin to diminish as the carbonate decomposes. Observed IR bands for all samples studied before heating are listed, together with assignments in Table 5.
Our experiments therefore give clear temperature markers that can be looked for if portable ATR-IR spectrometers were to be used to investigate paints which have been heated to high temperatures in situ, for example, at a Scene of Crime. Our study is of a preliminary nature and ideally a much wider range of paints and different surfaces should be studied. Moreover, it should be noted that our laboratory-based study has focused on the effect of heat on paint samples. To extend this to the effect of fires at a scene of crime further experiments should be performed to assess the effect of other factors arising from the fire, for example, smoke contamination on IR measurements. However, these results are promising and we suggest that further investigation should be made as to the usefulness of such measurements in Scene of Crime investigation where it is important to ascertain the approximate temperature to which a painted surface has been heated. Portable ATR-IR, Raman, and XRF spectrometers are now available. Our studies have also shown much more potential for the use of ATR-IR rather than the other techniques for this specific application. Problems of fluorescence and sample degradation mean that Raman spectroscopy can never be universally applicable to a wide range of  paints seriously limiting its use in forensic applications. XRF gives elemental composition but does not give information on chemical changes upon heating which are forthcoming from ATR-IR spectroscopy. We also note that XRF measurements show much more interference from the substrate material as Xrays penetrate the sample much further than does IR radiation.
Our studies also show that ATR-IR investigation gives more unambiguous data than simple visual inspection. Visual inspection is prone to error given that paints typically darken in color, then lighten, and finally darken again upon heating, making esti-mates of the temperature to which they have been heated difficult. However, it should be noted that ATR-IR is not ideal for observing bands in the far-IR region below 600 cm-1 where metal oxide products will absorb.